Received: 21 June 2017 | Accepted: 30 January 2018 DOI: 10.1111/1365-2745.12956 RESEARCH ARTICLE Anti- herbivore defences and insect herbivory: Interactive effects of drought and tree neighbours Bastien Castagneyrol1 | Hervé Jactel1 | Xoaquín Moreira2 1BIOGECO, INRA, University of Bordeaux, Cestas, France Abstract 2Misión Biológica de Galicia (MBG-CSIC), 1. How much a plant is attacked by insect herbivores likely depends on its apparency Pontevedra, Galicia, Spain and ability to produce defensive traits, which may be modified by neighbouring Correspondence plants and abiotic conditions. Yet, how much the direct and trait-mediated effects Bastien Castagneyrol of neighbours on herbivory is modified by abiotic factors is still unknown. Email: [email protected] 2. By using a tree diversity experiment in SW France, we measured leaf insect her- Funding information bivory (chewers and miners), nutritional quality (water content, C/N ratio, sugar French Ministry of Ecology and starch content) and chemical defences (total polyphenolics and condensed Handling Editor: Matthew Heard tannins) on birch (Betula pendula) trees growing in monocultures and mixtures with oak, pine or both species. We alleviated water stress by irrigating trees in half of the plots while trees remained unwatered (i.e. drought-stressed) in the other half. 3. Overall, insect herbivory was higher among heterospecific neighbours, which cor- responds to associational susceptibility. Consistent with this finding, leaves had lower amount of anti-herbivore defences among heterospecific neighbours. In turn, insects caused more damage in drought-stressed conditions, but such effect was independent of leaf chemistry. We also found that the effect of tree species diversity on herbivory was contingent of drought conditions as associational sus- ceptibility only occurred in drought-stressed trees. The independent and interac- tive effects of neighbour diversity and irrigation on leaf herbivory remained significant after accounting for birch apparency and leaf chemistry, suggesting that unmeasured plant traits or some other mechanisms not associated with plant trait variation and apparency might be involved in the observed herbivory patterns. 4. Synthesis. By demonstrating that associational effects are contingent upon abiotic constraints, we bring new insights into our understanding of the mechanisms driv- ing diversity—resistance relationships across climatic gradients. KEYWORDS associational effects, Betula pendula, biodiversity, chewers, forest, insect herbivory, miners, ORPHEE experiment Journal of Ecology. 2018;106:2043–2057. wileyonlinelibrary.com/journal/jec © 2018 The Authors. Journal of Ecology | 2043 © 2018 British Ecological Society 2044 | Journal of Ecology CASTAGNEYROL et al. 1 | INTRODUCTION most part, but recent studies addressing the effect of abiotic factors on associational resistance suggest that there are important insights Ecological research conducted over the last two decades has shown to be gained from considering the two drivers together (Glassmire that plant species diversity has substantial effects on ecosystem et al., 2016; Grettenberger & Tooker, 2016; Kambach, Kühn, processes such as productivity (Cardinale et al., 2011), and also in- Castagneyrol, & Bruelheide, 2016; Walter et al., 2011). Indeed, it is fluence arthropod community structure and overall species rich- increasingly acknowledged that plant neighbourhood composition ness at higher trophic levels (Barbosa et al., 2009; Cardinale et al., can modulate the way plants respond to water stress (Forrester, 2011; Moreira, Abdala- Roberts, Rasmann, Castagneyrol, & Mooney, Theiveyanathan, Collopy, & Marcar, 2010; Klaus et al., 2016; Otieno 2016; Scherber et al., 2010). In particular, it is well- established that et al., 2012). For instance, in accordance with the stress gradient a greater diversity of plant species reduces herbivore attack by de- hypothesis predicting reduced competition and increased facilita- creasing the concentration (i.e. number of individuals) and frequency tion under high stress (Callaway & Walker, 1997), Forey et al. (2016) (i.e. relative abundance) of herbivores’ hosts (“the resource concen- found that beech trees growing in mixed stands increase soil nutri- tration hypothesis,” Hambäck, Inouye, Andersson, & Underwood, ent acquisition in comparison to those trees growing in monospecific 2014; Root, 1973; Underwood, Inouye, & Hambäck, 2014), or by stands, in particular in water- stressed soils. It is therefore likely that decreasing plant apparency through an interference with physical the effects of plant diversity on leaf quality and resulting herbiv- and chemical cues used by insect herbivores to detect and reach ory are modified by drought (Grettenberger & Tooker, 2016; Walter their host plant (Castagneyrol, Giffard, Péré, & Jactel, 2013; Jactel, et al., 2011). Birgersson, Andersson, & Schlyter, 2011). In addition, a growing num- In the present study, we tested for independent and interactive ber of studies has reported that plant species diversity can indirectly effects of tree species diversity and drought on insect leaf herbiv- affect herbivory on a focal plant by modifying its nutritional quality ory (miners and chewers) and leaf chemical traits associated with and anti- herbivore defences (e.g. physical traits and secondary me- defence and nutritional quality (phenolic compounds, C:N ratio, tabolites, Kostenko, Mulder, Courbois, & Bezemer, 2017; Moreira, water content and carbohydrates) in silver birch trees (Betula pen- Abdala- Roberts, Parra- Tabla, & Mooney, 2014; Mraja, Unsicker, dula, Betulacae), a common broadleaved species in the study area. In Reichelt, Gershenzon, & Roscher, 2011), independently of resource addition, we investigated if the measured tree traits, as well as tree concentration and host–plant apparency. However, observed pat- apparency (measured as how much individual birches were taller terns are highly inconsistent and the mechanisms underlying such than their neighbours, e.g. Castagneyrol et al., 2013), underlied the patterns are poorly understood. For example, some studies have effects of neighbourhood diversity and drought on leaf herbivory reported increased investment in anti- herbivore defences among in birch trees. To test for these effects, we used a factorial field ex- heterospecific neighbours (Moreira et al., 2014; Mraja et al., 2011), periment where we manipulated both host–plant species diversity while others have reported the opposite (Glassmire et al., 2016; (three levels: birch monocultures, two- species mixtures associating Kostenko et al., 2017), generating a negative (Glassmire et al., 2016) birch with the pedunculate oak Quercus robur or maritime pine Pinus or a neutral effect (Kos, Bukovinszky, Mulder, & Bezemer, 2015; pinaster, and three- species mixture with pedunculate oak, the mar- Moreira et al., 2014) on insect herbivores. itime pine and birch) and water availability (two levels: irrigated vs. A likely source of inconsistencies among studies addressing plant non- irrigated). By addressing these goals, this study builds towards diversity effects on plant–herbivore interactions is the lack of con- a better mechanistic understanding of the combined effects of plant sideration of abiotic factors. It is well known that abiotic stresses species diversity and abiotic conditions on plant–herbivore interac- such as drought directly and indirectly alter plant–herbivore inter- tions and plant traits associated with resistance to herbivory. actions (Huberty & Denno, 2004; Jactel et al., 2012; Walter et al., 2011). Herbivory consumption rates and fitness, as well as insect 2 | MATERIALS AND METHODS herbivore performance, may drastically increase with increased drought stress because drought may indirectly modify plant nutri- 2.1 | Natural history tional quality and investment in anti- herbivore defences (Huberty & Denno, 2004; Jactel et al., 2012; Walter et al., 2011; White, 1974, Silver birch (B. pendula, Roth, Betulacae) is a broadleaved woody 2009). In this sense, the plant stress hypothesis posits that abiotic species with wide climatic and edaphic tolerance. Its distribution stresses on plants such as lack of water enhance the nutritional qual- range covers all Europe (Atkinson, 1992). Being a common pioneer ity of foliage for insect herbivores by altering biochemical source– species in northern and western Europe, the silver birch is naturally sink relationships and foliar chemistry (Walter et al., 2011; White, present in the study area where it is frequently found in association 1974, 2009). For instance, Ximénez- Embún, Ortego, and Castañera with maritime pine and various oak species. Betula pendula is intoler- (2016) reported an increase in amino acid and sugar concentrations ant to drought (Atkinson, 1992) which causes premature leaf drop in in leaves of drought- stressed tomato plants, resulting in increased summer (B. Castagneyrol, pers. obs.). Betula pendula is attacked by performance of the mite Tetranychus evansi. a large community of insect herbivores, especially leaf-chewers and To date, theory on plant diversity and drought effects on plant– miners such as Operophtera fagata, Hemichroa crocea or Eriocrania herbivore interactions have been developed independently for the spp. (Atkinson, 1992; Kunca, Csoka, & Zubrik, 2013). Even at low CASTAGNEYROL et al. Journal of Ecolog y | 2045 damage levels, these herbivores can cause significant growth loss on P. pinaster), with an additional replicate